Ανάλυση και πρόβλεψη χρονοσειρών στην τουριστική βιομηχανία
Date Issued
June 4, 2024
Type
Πτυχιακή Εργασία
Abstract
The dissertation critically examines the pivotal role of forecasting in the tourism and aviation industries. Over recent years, major software companies have dedicated efforts to providing forecasts, assisting businesses and organizations in making informed short-term and long-term decisions aimed at optimizing operational day-to-day activities and crucial long-term strategies. Beyond the realm of software companies, numerous organizations and academic researchers have developed their own tools, harnessing open-source resources like Python and other open-source libraries, as demonstrated in this dissertation.
The primary objectives of the dissertation revolve around deepening our understanding of time series forecasting applications. Specifically, the study addresses the challenges of forecasting during non-normal periods, exemplified by the impact of the COVID-19 pandemic on passenger arrivals at Athens International Airport. Additionally, the research seeks to compare traditional time series models like SARIMA (Seasonal AutoRegressive Integrated Moving Average) with modern deep learning models like LSTM (Long Short-Term Memory). This comparative analysis involves experimenting with optimization techniques, including grid search, autoARIMA, and rolling forecast origin.
The dissertation is structured into three main sections. The initial part encompasses the introduction and a comprehensive literature review. The second part delves into the experimental phase, which is divided into two segments. The first segment introduces models to a dataset without anomalies and attempts forecasting, while the second phase introduces anomalies into the training data. Finally, the dissertation concludes with findings derived from the experiments. Through this research, the author aims to provide valuable insights into the effectiveness of various forecasting models, particularly in the context of disruptive events like the COVID-19 pandemic.
Subjects